Pupillary dilations of mice performing a vibrotactile discrimination task reflect task engagement and response confidence

Abstract

Pupillometry, the measure of pupil size and reactivity, has been widely used to assess cognitive processes. As such, changes in pupil size have been shown to correlate with arousal, locomotion, cortical state and decision-making processes. In addition, pupillary responses have been linked to the activity of neuromodulatory systems that modulate attention and perception as the noradrenergic and cholinergic systems. Due to the extent of processes reflected by the pupil, we aimed at resolving pupillary responses in context of behavioral state and task performance while recording pupillary transients of mice performing a vibrotactile two-alternative forced choice task (2-AFC). We show that pre-stimulus pupil size differentiates between states of disengagement from task performance versus active engagement. In addition, when actively engaged, post-stimulus, pupillary dilations for correct responses are larger than for error responses with this difference reflecting response confidence. Importantly, in a delayed 2-AFC task version, we show that even though pupillary transients mainly reflect motor output or reward anticipation following the response of the animal, they also reflect animal decision confidence prior to its response. Finally, in a condition of passive engagement, when stimulus has no task relevance with reward provided automatically, pupillary dilations reflect stimulation and reward being reduced relative to a state of active engagement explained by shifts of attention from task variables. Our results provide further evidence for how pupillary dilations reflect cognitive processes in a task relevant context, showing that the pupil reflects response confidence and baseline pupil size encodes attentiveness rather than general arousal. Significance StatementFor the last 60 years, pupillometry has been used to study various cognitive processes. Among which are mental load, arousal and various decision related components, linking pupil dilations to underlying neuromodulatory systems. Our results provide extensive evidence that in addition to reflecting attentiveness under task performance, pupil dilations also reflect the confidence of the subject in his ensuing response. This confidence coding is overlaid within a more pronounced pupil dilation that reflects motor output or other post-decision components such that are related to the response itself but not to the decision. Our results also provide evidence how different behavioral states, imposed by task demands, modulate what the pupil is reflecting, presumably showing what the underlying cognitive network is coding for.

The Cortical States of Wakefulness

“…Task performance is higher during immobile, quiet behavior with high arousal than during lower level of arousal or high motor activity (McGinley et al, 2015b; de Gee et al, 2018; Neske and McCormick, 2018; van Kempen et al, 2018). These changes in brain state can occur spontaneously during the task or reflect overall changes in task engagement (Ganea et al, 2018; Jacobs et al, 2018), Moreover, they are well correlated to pupil size and could relate partly to fluctuations in the activity of locus coeruleus noradrenergic neurons in the brainstem (de Gee et al, 2017, 2018).…”

Neuronal correlates of the subjective experience of attention

“…In macaque monkeys, attention to the fingertips has been shown to increase the firing rate of neurons with corresponding receptive fields in the primary somatosensory cortex (Iriki et al, 1996). In rodents, pupil diameter was shown to be increased when rats were in a high behavioural state, defined by enhanced somatosensory detection, increased sensory evoked responses and lower population variability in the somatosensory cortex (Ganea et al, 2020; Lee & Margolis, 2016).…”

Measuring stimulus-evoked neurophysiological differentiation in distinct populations of neurons in mouse visual cortex

“…Locomotion and pupil diameter can be considered behavioral indications of engagement with the environment (Jacobs et al, 2018;Ganea et al, 2018;Bennett et al, 2013) and modulate neuronal activity in visual cortex (Dadarlat & Stryker, 2017;McGinley, David, et al, 2015;McGinley, Vinck, et al, 2015;Niell & Stryker, 2010;Polack et al, 2013;Reimer et al, 2014;Salkoff et al, 2020;Vinck et al, 2015). We found that in L2…”